Repos is a source control product co-optimized with our suite of fast, specialized coding models. It's git compatible, high throughput, and has built-in codebase retrieval.
Hey everyone 👋 Preston, Eitan, and Julian here from Relace.
Over the past year, we’ve shipped a suite of models that make AI coding agents fast:
The fastest apply model on the market (10k+ tok/s)
State-of-the-art embedding and reranker models for code retrieval that process huge codebases in seconds.
These models have powered tens of millions of requests from amazing companies like Lovable, Figma, Magic Patterns, and Orchids.
Now we’re taking it a step further. LLMs have proven they can write code — but scaling that code into production still needs better infrastructure.
We're starting with the foundation, source control.
Relace Repos is a new kind of source control system — built for AI agents, not humans. It’s fully git-compatible, designed for high-frequency read/write operations, and deeply integrates with Relace models.
Repos automatically indexes your codebase with our embedding and reranker models. That means your agents can find relevant files instantly instead of crawling the entire project.
This is just the start. Over the next few months we'll be releasing our suite of specialized agents for tasks like quickly searching codebases, automatically merging conflicts, and intelligently refactoring codebases.
These functions will be as easy as repo.search(), repo.merge(), and repo.refactor(). We believe models need to be co-optimized with the best infrastructure to take vibe coding to the next level.
If you’re already building coding agents or want to integrate them into your product, we’d love to collaborate.
Thanks for checking us out — we can’t wait to see what you’ll build with Relace!
Hey everyone 👋 Preston, Eitan, and Julian here from Relace.
Over the past year, we’ve shipped a suite of models that make AI coding agents fast:
The fastest apply model on the market (10k+ tok/s)
State-of-the-art embedding and reranker models for code retrieval that process huge codebases in seconds.
These models have powered tens of millions of requests from amazing companies like Lovable, Figma, Magic Patterns, and Orchids.
Now we’re taking it a step further. LLMs have proven they can write code — but scaling that code into production still needs better infrastructure.
We're starting with the foundation, source control.
Relace Repos is a new kind of source control system — built for AI agents, not humans.
It’s fully git-compatible, designed for high-frequency read/write operations, and deeply integrates with Relace models.
Repos automatically indexes your codebase with our embedding and reranker models. That means your agents can find relevant files instantly instead of crawling the entire project.
This is just the start. Over the next few months we'll be releasing our suite of specialized agents for tasks like quickly searching codebases, automatically merging conflicts, and intelligently refactoring codebases.
These functions will be as easy as repo.search(), repo.merge(), and repo.refactor(). We believe models need to be co-optimized with the best infrastructure to take vibe coding to the next level.
If you’re already building coding agents or want to integrate them into your product, we’d love to collaborate.
Thanks for checking us out — we can’t wait to see what you’ll build with Relace!